Spaces:
Sleeping
Sleeping
feat(hchat): add QwQ to prod config (#1598)
Browse files* feat(hchat): add QwQ to prod config
* fix: change context to 16k
- chart/env/prod.yaml +17 -0
chart/env/prod.yaml
CHANGED
@@ -137,6 +137,23 @@ envVars:
|
|
137 |
}
|
138 |
]
|
139 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
140 |
{
|
141 |
"name": "nvidia/Llama-3.1-Nemotron-70B-Instruct-HF",
|
142 |
"tokenizer": "nvidia/Llama-3.1-Nemotron-70B-Instruct-HF",
|
|
|
137 |
}
|
138 |
]
|
139 |
},
|
140 |
+
{
|
141 |
+
"name": "Qwen/QwQ-32B-Preview",
|
142 |
+
"preprompt": "You are a helpful and harmless assistant. You are Qwen developed by Alibaba. You should think step-by-step.",
|
143 |
+
"modelUrl": "https://huggingface.co/Qwen/QwQ-32B-Preview",
|
144 |
+
"websiteUrl": "https://qwenlm.github.io/blog/qwq-32b-preview/",
|
145 |
+
"logoUrl": "https://huggingface.co/datasets/huggingchat/models-logo/resolve/main/qwen-logo.png",
|
146 |
+
"description": "QwQ is an experiment model from the Qwen Team with advanced reasoning capabilities.",
|
147 |
+
"parameters": {
|
148 |
+
"stop": ["<|im_end|>"],
|
149 |
+
"truncate": 12288,
|
150 |
+
"max_new_tokens": 4096,
|
151 |
+
"temperature": 0.7,
|
152 |
+
"top_k": 20,
|
153 |
+
"top_p": 0.8,
|
154 |
+
"repetition_penalty": 1.05
|
155 |
+
}
|
156 |
+
},
|
157 |
{
|
158 |
"name": "nvidia/Llama-3.1-Nemotron-70B-Instruct-HF",
|
159 |
"tokenizer": "nvidia/Llama-3.1-Nemotron-70B-Instruct-HF",
|